Live TV buffer/recording location + HDD life span

A place to talk about GPUs/Motherboards/CPUs/Cases/Remotes, etc.
wyerock

Posts: 6
Joined: Tue Oct 08, 2013 9:15 pm
Location:

HTPC Specs: Show details

#21

Post by wyerock » Thu Oct 10, 2013 9:30 pm

I have an Antec P280 with the default case fans pulling air out of the back and top of the case, with nothing directly on the hard drives, although the air is pulled from the front across the drives. I generally leave all the fans on low, and depending on usage the drives vary from 30-35C (as reported by HD Tune).

I was able to use DriveImage XML to copy the second drive and recover the data, but without knowing how much of it corrupted. Since its all transport streams, the errors aren't a big deal. Basically, the drive is not completely dead like the first one was, so I'm more likely to suspect a problem with my MB or cabling than WMC or temperature, especially since the other drives appear to be fine. (or coincidence)

And since you brought up the Google study:
While they found low temps associated with high failures, they found that the correlation between temperature in general and failure is not statistically significant. The reason it appears that low temps are bad, but doesn't show statistical correlation is because the sample size for low and high temperature drives is so small compared to drives at normal temperatures. I used Figure 4 to ROUGHLY calculate: only a tiny fraction of Google's drives reported lower than 20C; about 5% below 25C; 20% from 25-29C, 27% for 30-34C, 27% for 35-39C, 15% for 40-44; 4% for 45-47C; 1% over that.

Basically, 33C is the most common temperature for Google's drives and 65% run from 28-39C. Further, we know from many other studies that heat cycling is much more important than average temperature, and Google's data is for drives that remain spun up for their entire lives. So I would definitely not use Google's 6 year old study of drives that topped out at 400GB to suggest upping a drive temperature to 40C.

Also note, they found that activity did not have an impact on failure rates, so that too suggests WMC is not what killed my drives.

wyerock

Posts: 6
Joined: Tue Oct 08, 2013 9:15 pm
Location:

HTPC Specs: Show details

#22

Post by wyerock » Sun Oct 13, 2013 4:03 pm

Follow up: The warranty on my 2TB Samsung was extended when Seagate took them over, so thanks for that recommendation STC. It failed the short and long diagnostics in SeaTools, so its going back. The 500 GB works again after a long reformat, and SMART doesn't show any reallocated sectors or any other warning signs, so I'm back to using that--I'll probably go with the RAM disk suggestion when RAM prices come back down.

Post Reply